Segmentation of retinal blood vessels by means of 2D Gabor wavelet and fuzzy mathematical morphology

Retinopathy is the most prevalent visual deficiency in diabetic individuals, which originates from progressive growth of new vessels and leads to substantial changes in the structure/density of vessels in retina. Due to the low contrast of vessels in digital images acquired from retina, vessels are not easily visible by eye; therefore, it is desired to develop automatic techniques for isolating blood vessels. This can not only assist the expert in the examination of the structure of blood vessels, but also can be considered as an initial step for designing fully automatic methods of diagnosing the pathology. However, low contrast between the background and retinal blood vessels, makes their segmentation challenging. Yet, much study has been concerned with this obstacle, and plenty of approaches have been suggested. Here we represent that combining Gabor wavelet as an efficient technique for removing background (fundus); and fuzzy mathematical morphology as a segmentation method, leads to remarkable improvement in segmentation process. Therefore, the proposed pipeline can be considered as an effective approach to segregate vessel from non-vessel tissues in retina.